Kubernetes Troubleshoot

Overview

Kubernetes Troubleshoot is a Model Context Protocol server that provides read-only access to Kubernetes diagnostics including pods, logs, events, workloads, services, persistent volume claims, and nodes. By implementing the MCP standard, it lets AI assistants like Claude query your cluster as native tools rather than requiring manual context switching between applications.

The server runs locally via the Docker image docker.io/yawgyamfiprem32/k8s-troubleshoot-mcp:1.0.0, communicating over stdio. You'll need to set the KUBECONFIG and ALLOWED_NAMESPACES environment variables to connect it to your cluster. Version 1.0.0 is released under the MIT license with full source code available at NanaGyamfiPrempeh30/k8s-troubleshoot-mcp on GitHub.

Kubernetes Troubleshoot is listed in the Cloud Service category on mcp.site and works with any MCP client including Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, and Zed. If you maintain this server, please claim this listing to verify ownership and keep information current.

Installation

// claude_desktop_config.json
{
  "mcpServers": {
    "kubernetes_troubleshoot": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "docker.io/yawgyamfiprem32/k8s-troubleshoot-mcp:1.0.0"
      ],
      "env": {
        "KUBECONFIG": "<kubeconfig>",
        "ALLOWED_NAMESPACES": "<allowed-namespaces>"
      }
    }
  }
}

Requires KUBECONFIG, ALLOWED_NAMESPACES